Emotion Decoding: An Extensive Examination of Electroencephalogram Signals Using Explainable Machine Learning

Emotion plays a substantial impact on an individual's mental processes and social interactions. It functions as a connection between an individual's emotions and their behaviors, or it may be stated that it occasionally impacts one's life choices. Given the individual variations in emotional patterns and their manifestations, the investigation of these phenomena should rely on methodologies that are applicable across diverse populations. In order to improve accuracy and pick specific features, the process of emotion detection utilizing brain waves or EEG data necessitates the utilization of effective signal processing techniques. Researchers have been working on several methods of human-machine interface technologies for a considerable period. In recent years, they have achieved significant progress in the automated comprehension of emotions through brain signals. Our research involved the classification and testing of three emotional states using EEG signals collected from the widely accessible EEG Brainwave Dataset: Feeling Emotions from kaggle, utilizing seven machine learning techniques. This study presented a methodology that employed machine learning to identify emotions using the EEG Brainwave dataset. The study also conducted a thorough evaluation of several machine learning algorithms to assess their overall effectiveness. The study employed the Local Interpretable Model-Agnostic Explanations (LIME) to generate interpretable predictions and gain insights into the elements that influence the model's predictions. Despite the interpretability of the ML models, the XGBoost classifier had the greatest accuracy of 99% in recognition.